By John Wheeler | Wheeler Intelligent Systems
A plant manager asks for a spec sheet, a test run using the plant’s own cartons, and a quote for a new case sealer. The salesperson says, “I’ll get back to you Thursday.” Then the next call starts.
By Friday, the spec sheet has gone out, the quote is still waiting on installation costs, and the test request is buried in a notebook. The plant manager needs to know whether to set aside time for the test. If you own a small business, you probably know how this happens. Each step is simple. Keeping track of all of them across a busy week is the hard part.
On November 17, OpenAI introduced ChatGPT for Sales. The announcement caught my attention because it points to a practical use for AI in a small business. What if the work you did before the call, the promises you made during it, and the follow-up you owe afterward could all carry forward? OpenAI’s ChatGPT for Sales announcement
What OpenAI introduced
ChatGPT for Sales includes a Sales plugin with ready-made workflows and more than 30 new or expanded partner plugins. OpenAI describes using connected tools such as HubSpot or Salesforce for account history and Gong or Granola for call notes. That gives a salesperson a way to bring those records into ChatGPT when preparing for a meeting or writing a follow-up. OpenAI says partner plugin and feature availability varies by plan, region, and workspace settings. OpenAI’s launch details and availability
If your team already keeps an account record in a CRM, that record can give AI a starting point. The call notes might explain why a plant needs to seal two carton sizes on the same line. An approved product sheet can tell the salesperson which machine models are designed to handle those sizes. A good meeting brief would bring those facts together and point out what still needs an answer.
For a small team, I’d start with the moment right after the call. Everyone knows what should happen then. The salesperson needs to record what the buyer asked for, confirm what was promised, and get the right people working on it. You can also check the result against the notes while the conversation is still fresh.
What this could look like at a packaging equipment supplier
Imagine a fictional equipment supplier selling a case sealer to a company that makes light fixtures. A case sealer tapes cartons shut as they move down the packing line. The plant manager needs one machine that can handle two carton sizes. She asks for a spec sheet, a test run with both sizes before next Wednesday’s project meeting, and a quote that includes installation.
The salesperson leaves the call with two commitments. He’ll send the spec sheet, and he’ll check whether the test can happen before the meeting. Installation pricing is still open. The service team needs to see how the new machine would fit into the existing line before it can estimate the work.
In a trial, the salesperson gives ChatGPT the call notes and the approved spec sheet. He asks for a short buyer email and a list of internal tasks. The email confirms that he’s sending the spec sheet and will check on the test. The task list asks the shop to confirm its schedule and what the buyer needs to send. It also flags the information the service team needs for the installation quote.
Now imagine the draft says, “We’ll have both carton sizes tested by Wednesday.” That sounds like a small change. To the plant manager, it’s a promise. She might put the results on the meeting agenda and ask her team to make a decision that day. If the shop can’t fit in the test, she has to change her plans. That one sentence would be enough for me to send the draft back for a correction.
There’s a second detail I’d check. A spec sheet may show that both sizes fall within the machine’s stated range. The test still has a purpose. It gives the buyer a chance to see how the machine handles the actual cartons coming off the line. I’d want the draft to keep that question open until the shop has results.
That’s where the sales process meets the rest of the business. The CRM might show what the buyer wants. The shop schedule tells you when a test can happen, and the service team knows what it takes to install the machine. If the AI has only the call notes, I’d expect it to tell me what needs checking and who can answer.
Could it help the owner see which deals need attention?
Picture a Monday sales meeting. The case sealer deal still shows an expected close date at the end of the month. The salesperson says the buyer liked the machine. The last email gives a more specific picture. The plant manager wants to see both carton sizes run before taking the purchase to her boss. Three weeks have passed. The shop hasn’t received the cartons, and there’s no test date on the calendar.
I’d want AI to bring that deal up and show me the email that explains why. A reasonable next step would be to ask when the buyer can send the cartons. Then the salesperson can add what they know from working with the buyer. Maybe the plant manager said by phone that one of the cartons is being redesigned. Once that detail is in the record, the owner has a better reason to revisit the close date.
It also gives the owner a way to help. If the buyer can ship the cartons this week, the owner might ask the shop to hold an opening. If the carton design is still a month out, the salesperson can agree on a date to check back. Either way, the team leaves the sales meeting knowing what it needs to do.
How I’d test it in a small business
I’d pick one salesperson and one repeatable task. For two weeks, test whether ChatGPT can prepare an accurate follow-up after a sales call. I’d start with made-up calls and product details so the team can practice catching mistakes. Include an easy case, one where the buyer asks for an unapproved discount, and one where the buyer suggests a date the salesperson never confirms.
I’d give the AI a simple instruction like this.
Draft a short email based on these call notes. Confirm only the commitments we made. List open questions and internal tasks separately. Add an owner or due date only when the notes support it. Flag any price, delivery date, or product claim that needs approval. Show me the part of the notes that supports each commitment.
Then I’d time the whole job, including the review and the fixes. If an AI draft takes two minutes to make and eight minutes to check, record all ten. I’d also track missed commitments, promises added by mistake, and whether follow-up went out when the salesperson said it would. Over a few weeks, those measures tell you more about this task than a change in total sales would.
For the first live trial, I’d have the salesperson approve and send the email and record the agreed next step in the CRM. I’d compare the results with a recent set of calls handled the usual way. If the team spends less time and keeps its promises more consistently, I’d try a second task, such as preparing a one-page brief before a buyer meeting.
What I’m watching
When I test AI on everyday business tasks at Wheeler Intelligent Systems, I want to see whether it handles the details people count on. Sales is a good place to look. A buyer who asked to test two carton sizes should hear back about both. A quote waiting on installation costs should have someone assigned to it. A requested test date should show up as a request until the shop confirms it.
Think about the last sales call your company handled. Could someone who missed the call tell what the buyer asked for, what your team promised, and who owes the next step? If that takes a search through email, notes, and someone’s memory, you’ve found a place to test ChatGPT for Sales.
